For the problem of point source forming prescribed irradiance, a new, to the best of our knowledge, method-variable separation mapping method is presented, which establishes separately the correspondence between variables on the light source and the target plane. The role played by the optical surfaces is then to redirect the light rays to their corresponding target points. The surface of the lens is determined by first calculating the surface points and then their normal vectors. Considering that normal deviations are produced in the surface construction process, a normal deviation control method is also presented to restrict the deviation. With this normal deviation control method, discontinuities are introduced onto the lens surface. From these mapping and normal control methods, a fast and efficient algorithm has been developed for several prescribed irradiance problems with simple nonrotational shape of the illuminated region.

No abstract

A new method is proposed to achieve high illuminance and luminance uniformity of the road surface in LED road lighting. Based on the reflection properties of the road surface, the illuminance and luminance are analyzed simultaneously with the least-square method; meanwhile, energy efficiency and glare requirements are considered. Through the analysis and calculations, the optimal light distribution of a luminaire is obtained, and then a freeform lens with this light distribution is designed. For a 2-lane C1 class road illuminated by LED luminaires mounted with these lenses, the overall illuminance and luminance uniformity on the road surface can reach over 0.9 and 0.85, respectively, and the glare factors less than 10%.

The availability of the large-scale labeled 3D poses in the Human3.6M dataset plays an important role in advancing the algorithms for 3D human pose estimation from a still image. We observe that recent innovation in this area mainly focuses on new techniques that explicitly address the generalization issue when using this dataset, because this database is constructed in a highly controlled environment with limited human subjects and background variations. Despite such efforts, we can show that the results of the current methods are still error-prone especially when tested against the images taken in-the-wild. In this paper, we aim to tackle this problem from a different perspective. We propose a principled approach to generate high quality 3D pose ground truth given any in-the-wild image with a person inside. We achieve this by first devising a novel stereo inspired neural network to directly map any 2D pose to high quality 3D counterpart. We then perform a carefully designed geometric searching scheme to further refine the joints. Based on this scheme, we build a large-scale dataset with 400,000 in-the-wild images and their corresponding 3D pose ground truth. This enables the training of a high quality neural network model, without specialized training scheme and auxiliary loss function, which performs favorably against the state-of-the-art 3D pose estimation methods. We also evaluate the generalization ability of our model both quantitatively and qualitatively. Results show that our approach convincingly outperforms the previous methods. We make our dataset and code publicly available. 1